System and Method for Counting Poultry or the Like in a Caging Machine

The present invention relates to the sector of rearing poultry animals, such as chickens, guinea fowl, turkeys or similar animals.

It is known that most of the poultry animals (henceforth referred to more simply as “poultry”) intended for slaughtering are currently reared in sheds or the like. Thousands of poultry are reared in the sheds using very advanced rearing techniques. Intensive rearing systems of this type allow the farm managers to achieve significant economies-of-scale which result in economic advantages for consumers who acquire cuts of meat from animals reared using these techniques. Moreover, these types of intensive rearing systems result in a substantial uniformity of the reared animals. After a certain predefined period of time the reared poultry reach the correct weight and are ready to be slaughtered.

At the end of the rearing period, the live animals must be removed, without causing them injury or unacceptable suffering, from the shed where they were reared and then taken to the next stages in the supply chain, i.e. typically transported to the slaughterhouse.

Machines for removing the poultry from the rearing sheds and arranging them inside cages are known. The cages filled with poultry are then loaded onto lorries or their trailers and transported to the slaughterhouse.

The removal (or caging) operations are typically performed using special machines, commonly known as “poultry loading machines” (for example “chicken loading machines” in the case where the poultry consists of chickens) or “caging machines”.

At present, operation of the caging machines is based on the weight of the animals loaded into the cages, with a standard-statistically determined-weight of a single animal, for example a chicken, being set in the caging machine; the cage being filled is weighed by means of a weighing system positioned underneath the cage itself. Whenever a new empty cage reaches the loading/filling position, the weight is zeroed in order to determine the tare weight of the cage itself; in the caging machine the overall weight of the animals to be loaded in each cage is set and, when the set weight for each cage is reached, the caging machine stops, the cage is considered filled and the next cage is then filled. Determination of the quantity (number) of animals, for example chickens, which are placed in each cage is therefore performed in an indirect manner, based on a standard statistical weight of the single animal; and the overall weight of the cage.

The Applicant has noted that the known systems for removing the poultry from the rearing sheds and caging them have major drawbacks.

In particular the detection of the overall weight of the animals removed from the rearing shed is performed inside the cage itself. Therefore determination of the filled state of the cage occurs always with a delay; for example, if a cage must be filled with 100 animals and, by means of weighing, the machine determines that at a certain moment a cage weight corresponding to 99 units of standard weight has been reached, it is difficult to ensure that the machine will stop exactly when 100 animals have been loaded; in fact, if on the conveyor belt there are several animals bunched together, it is likely that more animals than the last animal needed to reach the total of 100 will enter into the cage. This therefore results in cages which are generally filled with a number of animals different from that required.

Another inherent problem in the indirect determination of the quantity (number) of animals, for example chickens, which are present in each single cage, based on a standard statistical weight of the single animal and on the overall weight of the cage, i.e. by means of division of the overall weight of the cage by the average statistical weight of the said animals, is that not all the animals are the same, the weight of each single animal is different from that of other animals, and therefore the indirectly calculated number is not correct. However, the slaughterhouse, when informing the rearing farm about the loading data, will request the consignment of a specific number of animals to be slaughtered, and not a number corresponding to their overall weight.

Furthermore, in the event of checks carried out by the authorities during transportation, for example in connection with animal protection rights, the only way for them to check compliance with these rights is by counting the number of animals in order to determine whether the filling coefficient per square centimetre is being respected.

Further problems also arise from the fact that many slaughterhouse companies, in particular those abroad, pay the rearing farmers for the animals “per head” and not on a weight basis, and likewise the teams of operators responsible for loading the reared poultry are paid according to the number of animals loaded, and not the overall weight.

In view of the above, the Applicant has devised a system and a method for counting poultry or the like in a caging machine which are able to overcome the drawbacks associated with the known solutions.

One aspect of the present invention relates to a counting system for counting poultry in a caging machine configured to convey poultry on a conveyor belt in order to introduce them into a cage.

The counting system comprises an acquisition group for acquiring a sequence of images of the same portion of said conveyor belt during successive time instants.

The counting system further comprises a processing apparatus.

The processing apparatus comprises an identification unit configured to process the sequence of images in order to identify the presence of said poultry in each image.

The processing apparatus comprises a tracking unit configured to calculate poultry tracks based on the identified poultry presence.

Each poultry track indicates the path followed by a respective bird along the conveyor belt.

The processing apparatus comprises a counting unit configured to calculate a count indicating a number of birds that have reached the cage on the basis of the calculated tracks.

The processing apparatus comprises an output unit for outputting the calculated count for control of the caging machine.

According to an embodiment of the present invention, said identification unit comprises an object detection unit which implements a neural network model trained to optimize the ability of said neutral network model to recognize the poultry presence in images.

According to an embodiment of the present invention, said neural network model is trained with a set of training images comprising training images depicting poultry.

According to an embodiment of the present invention, at least a portion of said training images depicting poultry have been subjected to one or more of the following data augmenting operations:

According to an embodiment of the present invention, the identification unit and the tracking unit are configured to operate asynchronously and independently to each other.

According to an embodiment of the present invention, said identification unit comprises an object detection unit configured to calculate, for each poultry presence identified in an image, coordinates of a corresponding bounding box indicating a position and a space occupied by said poultry presence in said image.

According to an embodiment of the present invention, said tracking unit implements a tracking algorithm configured to calculate each poultry track corresponding to a bird based on the evolution over time of the coordinates of the bounding boxes calculated for the presence of said bird in the images of said image sequence.

According to an embodiment of the present invention, said counting unit is configured to update a counter whenever the coordinates of a bounding box of a poultry track have reached or exceeded threshold values indicating that the corresponding bird has reached the cage.

According to an embodiment of the present invention, said threshold values correspond to coordinates of a line in said image.

According to an embodiment of the present invention, said threshold values correspond to coordinates of a bidimensional region in said image.

According to an embodiment of the present invention, said acquisition group and said processing apparatus are part of a single embedded hardware unit.

According to an embodiment of the present invention, said counting system further comprises a buffer configured to temporarily store one or more images of said sequence of images.

According to an embodiment of the present invention, the identification unit is configured to process one of the images stored in the buffer to identify the presence of said poultry.

According to an embodiment of the present invention, said acquisition group comprises a camera or a video camera which captures said portion of said conveyor belt.

According to an embodiment of the present invention, said acquisition group comprises an illuminating device for illuminating said portion of said conveyor belt.

According to an embodiment of the present invention, said camera or video camera is an infrared camera or video camera, and said illuminating device is an infrared illuminating device.

According to an embodiment of the present invention, said portion of said conveyor belt comprises at least one portion of the conveyor belt before the entrance to said cage.

Another aspect of the present invention relates to a caging machine.

The caging machine comprises a conveyor belt for conveying poultry towards a cage and the counting system.

According to an embodiment of the present invention, the caging machine further comprises a control unit in communication with the output unit of the counting system, wherein said control device is configured to stop the conveyor belt or to direct the conveyor belt towards another cage when the calculated count has reached a value corresponding to a predefined target value of poultry.

A further aspect of the present invention relates to a method for counting poultry in a caging machine configured to convey poultry on a conveyor belt in order to introduce them into a cage.

The method comprises acquiring a sequence of images of a same portion of said conveyor belt during successive time instants.

The method comprises processing the sequence of images in order to identify the presence of said poultry in each image.

The method comprises calculating poultry tracks on the basis of the poultry presence identified, each poultry track indicating the path followed by a respective bird along the conveyor belt.

The method comprises calculating a count indicating a number of birds that have reached the cage on the basis of the calculated tracks.

The method comprises outputting the calculated count for control of the caging machine.

With reference to the attached figures, in which identical reference numbers refer to identical or functionally equivalent parts,is an axonometric view of a caging machinefor caging poultry in cagesto which the concepts in accordance with embodiments of the present invention may be applied.

According to an embodiment of the present invention, the caging machinecomprises a rear sectionfor collecting the poultry, a central conveying and raising section, and a front caging section.

According to an embodiment of the present invention, the birds are collected from the ground at the rear section.

According to an embodiment of the present invention, the rear sectioncomprises a pick-up conveyor beltextending along the longitudinal direction of the caging machine.

In accordance with an embodiment of the present invention, the pick-up conveyor beltreceives the birds from the ground and transfers them to the central section. In accordance with an example of embodiment of the present invention, the pick-up conveyor beltcomprises a belt endlessly wound around a drive roller and an idle roller. Similar considerations are applicable also to cases where the pick-up conveyor belthas a different structure.